1. Field of the Invention
The present invention relates to a charge transfer apparatus for transferring charges, and an image pickup apparatus using the same.
2. Related Background Art
Conventional charge transfer elements are a CCD (Charge Coupled Device) and CSD (Charge Sweep Device; Japanese Patent Publication No. 63-38866 and Japanese Laid-Open Patent Application No. 2-63314), and are mainly applied to solid-state image pickup elements. These charge transfer elements consist of MOS diodes formed on a semiconductor, and receive signal charges in depletion layers formed near the semiconductor interface by controlling the gate electrode potentials of the MOSs. The CCD and CSD adopt different transfer methods. The CCD is constituted by forming multistage MOS diodes, a potential well is formed on each stage, and a plurality of independent signal charges are assigned to a plurality of potential wells. The gate electrode potential of MOS on the respective stage is sequentially changed to sequentially move the position of the potential well, and the signal charges are simultaneously transferred accordingly. The CSD is multistage MOS diodes or a single MOS diode. One CSD transfer path, which forms one potential well, receives only one signal charge, and thus the maximum transferable charge is large. Charges are transferred by sequentially changing the gate electrode potentials of MOSs and sequentially moving the potential barrier position in the potential well.
Noise generating during transfer in the charge transfer element is mainly a dark current generating from a semiconductor substrate or MOS interface. Both the CCD and CSD can transfer signal charges with small noise.
However, the CCD or CSD used in a solid-state image pickup element suffers the following problem.
As for an interline CCD which is most widely used among CCD solid-state image pickup elements, optical signal detection photodiodes are two-dimensionally arrayed, and signal charge transfer CCDs (vertical CCDs) are interposed between the photodiode columns. If the photodiode area is increased to increase the sensitivity in this arrangement, the CCD area must be decreased. All of the signal charges of photodiodes on one column are transferred at once to a plurality of potential wells in the vertical CCD and then transferred. The maximum charge amount per photodiode that can be transferred by the CCD is proportional to the area of one stage of MOS diodes which form one potential well. Thus, if the CCD area is decreased, the acceptable maximum charge amount is restricted. The processible maximum signal charge amount per photodiode determines the dynamic range of an image pickup apparatus. As for an interline CCD, a high-sensitivity design and a wide-dynamic-range design are contradictory to each other.
In the arrangement of a CSD solid-state image pickup element, similar to the interline CCD, CSD charge transfer paths are interposed between photodiode columns. In this case, the entire CSD region of one column receives the signal charge of one photodiode. Even if, therefore, the CSD transfer path is narrowed, the acceptable maximum charge amount is not actually restricted, and the processible maximum charge amount is determined by the maximum accumulation charge amount of one photodiode. The CSD solid-state image pickup element can simultaneously realize high sensitivity and wide dynamic range by increasing the photodiode area. However, the CSD can transfer only a set of signal charges at once, its operation is inevitably line-sequential driving, and the signal charges of photodiodes on a row selected for signal transfer must be transferred at high speed. Particularly in a movie image pickup apparatus, the signal read-out time of one row is determined by a standard, and the signal charges of a photodiode must be transferred to the output terminal of the CSD transfer path within this time.
A CSD charge transfer method for solving this problem will be explained with reference to FIGS. 1A to 1E. FIG. 1A is a schematic sectional view of a CSD, and FIGS. 1B to 1E are views of the potentials for explaining the transfer method. In FIG. 1A, a semiconductor substrate 1 is a p-type substrate. The semiconductor substrate 1, an insulating layer 2, and a gate electrode 3 made of polysilicon or the like form a MOS diode. Terminals φ1, φ2, φ3, φ4, and φ5 for supplying potentials to the electrode 3 supply potentials at positions apart from each other at a certain interval. A gate electrode 4 is made of polysilicon or the like. The semiconductor substrate 1, insulating layer 2, and gate electrode 4 form a MOS diode. A terminal φs supplies a potential to the electrode 4.
FIGS. 1B to 1E show the potentials of the charge transfer path, i.e., near the semiconductor interface below the MOS diode, and show transfer of a signal charge Qsig. In this case, the signal charge carriers to be transferred are electrons. In FIG. 1B, the terminals φ2, φ3, φ4, and φ5 are at potentials high enough to deplete the interface of the semiconductor substrate 1 below the electrode 3. At this time, the terminal φ1 is set at a lower potential than the terminal φ2. A potential gradient is generated between the terminals φ1 and φ2 of the electrode 3, and thus a potential gradient is generated in the transfer path below them. In this case, the electrons drift in the X direction in FIG. 1A. As the location where the electrode 3 gives a potential gradient sequentially shifts, as shown in FIGS. 1C to 1E, the signal charge drifts. The terminal φs applies a higher potential than that of the terminal φ5 to the electrode 4 in contact with the end of the electrode 3. A deep potential well is formed at the semiconductor interface below the electrode 4, and signal charges are finally collected to this well.
The CSD using this potential gradient can ensure a higher motion speed of signal charges than the CCD transferring charges mainly by charge diffusion. However, the CSD is difficult to realize high-speed charge transfer required for an image pickup apparatus such as a movie camera owing to the following problems.
(1) Since the capacitance between the MOS electrode and the semiconductor interface and that between the semiconductor interface and the semiconductor bulk are connected in series, only a value obtained by multiplying the potential gradient supplied to the MOS electrode by the capacitance division ratio is applied to the semiconductor interface. This capacitance division ratio is generally as low as about 0.1 to 0.3.
(2) Since the potential gradient supplied to the MOS electrode introduces an ohmic current through this portion, a large potential gradient is difficult to apply.
(3) The charge mobility at the semiconductor interface serving as a transfer path is lower than that in the semiconductor bulk.
The CSD is widely used in an infrared image pickup apparatus which must process a large amount of signal charges, but is difficult due to those problems to use in an image pickup apparatus for a visible region with a high signal read-out speed.